Zero-Shot Learning (ZSL)

A machine learning paradigm where an AI model can make accurate predictions on tasks it has never encountered during training.

Instead of learning from examples, the model uses prior knowledge and inference to generalize across different domains.

In AI agents, ZSL allows for flexibility in dealing with unforeseen scenarios without requiring new training data.